Quote:Original post by Alpha_ProgDes
Quote:Original post by LessBread
Quote:Original post by Alpha_ProgDes
Quote:Original post by LessBread
#3 My understanding was that the current upper bound was somewhere above 2000. For example, Myopic Rhino has a rating of 2088.
If the switch to a percentage rating is made, existing members should start out with a rating based on their current rating, perhaps the percentage of their current rating relative to Myopic Rhino. As an example, that would give you a 95% rating (1987/2088) to start with.
Let's say for instance, iMalc suddenly gets a rating of 2634 and I stay at 1626. My rating would go down because of it. At first glance, it would be misleading because one would think I was being disruptive, unhelpful, or just a plain ass and got rated down. It's also misleading to myself if I or no one else has an actual rating to manipulate or base the percentage on.
...to start with
If the switch was made at this time, your percentage would start at 1626/2088. I chose Myopic Rhino for that example because (afaik) he has the highest rating. If iMalc's rating shot up to 2634 before that time giving him the highest rating, then his score would become the upper bound and your percentage would start at 1626/2634.
Well see my example was based on the fact, that the switch was done now and then iMalc shot up to 2634. Hence, my conclusion above.
It seems to me that your example was based on a hypothetical situation not a fact. More importantly though, it seems to me that you were misinterpreting my suggestion. I wasn't saying that percentage ratings should float in relation to the highest rating, only that current ratings should be carried over in someway
when the switch is made (ie. "to start with"). After the switch they would adjust according to whatever new mechanisms were put in place. In addition, I should have suggested that if there was a desire to level the playing field somewhat during the switch, then at least the current ratings should be factored into the new ratings in some way.
Quote:Original post by Alpha_ProgDes
Quote:Unidimensional ratings are limited in their explanatory power. Using the President as an example, he's rated low by people on the left because of Iraq and he's rated low by people on the right because of profligate government spending and a wishy washy immigration policy. His popularity number doesn't reflect the reasons it is what it is. To do that requires comparing responses to other polling questions. This is a problem similar to that of using left-right ideological distinctions when we here know from experience that a compass offers a superior means of identifying a persons ideological location.
From this perspective it seems that two ratings might be better as it would allow for a more precise coordination of members relative to other members. It might not be though because it might complicate the matter beyond usefulness.
I'll admit that my vocabulary overall is severely lacking. But that explanation was more confusing than explanatory to be honest.
But I will say this (because I believe I got the gist of what you said), the current system allows us to correctly assume the following: 1) the user rated based on what he/she said, 2) the user rating shows if the user has been helpful to the community, 3) if there's a high rating that user must be a valuable asset. Now this doesn't determine whether the user is extrememly funny, logical, or technical. It only determines that whatever that person does he/she are good at it and the community rewarded them for it.
The percentage system as you put it takes away hard numbers and instead has your rating as a ratio against the highest rating. Now your rating may go up if the user with the highest rating begins to be rated badly or you begin to be rated well. But no one knows because there's no numbers, just ratios. It just seems like another layer which hides needed information.
I hope my remark above clarified my initial suggestion. The highest rating would only play in at the switch over. It seems to me that a compass approach - that is two ratings, ie coordinates - would resolve some of the issue surrounding what kind of asset a member is or isn't. The question in that regard would be what each axis should measure. If old ratings were to be carried over than one axis would be related to the old rating, which is whether the member is "helpful and/or friendly" and the other would probably be relative to technical knowledge.
"I thought what I'd do was, I'd pretend I was one of those deaf-mutes." - the Laughing Man